摘要
本文研究了各总体服从多元正态分布 ,其未知参数的先验分布均为扩散先验分布时 ,如何利用待判样品的预报密度函数、构造后验概率比并据此对样品进行分类与判别 ;此方法并不需要假设各总体分布的协方差相同 ,而且在预试样本容量较小时仍然可行。
In this paper,the authors study a method on how to classify a sample into one of the several known normal populations in terms of posterior probability ratio established by the samples predictive density functions when the unknown parameters prior distributions are diffuse prior distribution.The method doesnt require the consistency of each populations covariance and it remains valid when the training sample size is not very large.
出处
《数理统计与管理》
CSSCI
北大核心
2003年第1期33-37,共5页
Journal of Applied Statistics and Management
